Atrial fibrillation (AF) is the most common heart rhythm disorder, affecting 2.2 million individuals in the United States alone, and is a major cause of morbidity and mortality. Current methods to eliminate AF with anti-arrhythmic drugs and ablation remain suboptimal, reflecting our current lack of understanding of the mechanisms for AF, and how they may differ for patients with presentations such as persistent or paroxysmal AF. This project tests the novel hypothesis that interaction of the dynamic tissue properties of repolarization and conduction with structural heterogeneities provides a direct mechanism for the initiation of human AF and its varying clinical patterns. This project builds upon published work and preliminary observations by our laboratory in patients. We have three specific Aims. 1) To determine whether dynamic tissue properties, including restitution of action potential duration, cause the initiation of atrial fibrillation;2) To determine whether the initiation of atrial fibrillation follows conduction block and reentry;3) To determine whether dynamic tissue properties are required to cause AF in computer models created specifically for each patient, then referenced back to observed AF. We will pursue these aims by acquiring high-resolution electrophysiologic and anatomic data at electrophysiologic study, by performing numerical analysis of activation in both atria, then by developing patient-specific computer models. The computer models that we will create in this project will be among the most detailed and clinically-relevant. This project is significant because it studies a novel mechanism for the development of atrial fibrillation in patients. This mechanism may serve as a method to predict the propensity for AF. Understanding this mechanism may also allow a more rational approach both to drug development and ablation therapy. The performance of this project in patients during electrophysiologic study will also allow its results to be translated directly to practice. Finally, our patient-specific computational models are clinically relevant, and will thus provide a resource for further hypothesis testing in AF.